Time precision on Linux and Windows

Unit tests were needed on a new project I’m working on and a few weeks ago I wrote some. Last week, a colleague wrote some more, and when he ran all of them, mine were failing. Same code base, up to date, same tests. On my machine they were still passing. We jumped on the situation and saw math.Floor was giving a result on my machine, which runs on Ubuntu, and another one on his, on Windows.

The code is larger, but I’ve extracted and adapted what’s needed:

package main

import (
	"time"
	"math"
)

func main() {
	datetime := time.Now().Add(time.Hour * 24 * 7 * 4 * 12 * 3)
	seconds := -1 * int(time.Now().Sub(datetime).Seconds())
	a := 29030400
	x := float64(seconds)/float64(a)

	println("input:", x, "floor:", math.Floor(x))
}

Result on Ubuntu:

input: +3.000000e+000 floor: +2.000000e+000

Result on Windows:

input: +3.000000e+000 floor: +3.000000e+000

Continue reading Time precision on Linux and Windows

Handling API errors

The past days I’ve practiced Go by writing a package for Apixu weather service. They have a a straightforward REST service to offer their weather information.

Their service also returns an error response if it can’t offer data based on your needs, if you use an invalid API key, or for other cases. Of course, the Go package should also return errors if the case.

No problem with returning data, but I had some issues handling errors. There can be general errors that have nothing to do with Apixu (but with the package internals) and errors returned by them. My first approach was to return three values for each API method:

Search(q string) (Search, ApixuError, error)

But it smelled right away. There had to be another way. Continue reading Handling API errors

Reloading Go apps automatically while developing

There are some ways to automatically reload Go apps while developing, to not have to manually stop your app, build it, run it. Recently I ran into this article about a nice tool called Fresh.

I wanted to start using live reload (or hot reload) for a project of mine, but it had a particularity which gave me troubles when I tried Fresh. My case was:

  • I had a multiple apps Git repository
  • The apps were sharing some packages
  • If I edited app1 and the packages it used, I didn’t want both app1 and app2 to be restarted, only app1 (similar for app2)

Continue reading Reloading Go apps automatically while developing

Counting number of items in a concurrent map

Lately I’ve been using Go’s concurrent map. And sometimes I needed to count the items I’ve stored in the map. I have ranged over the items and incremented a counter, and it was done:

package main

import (
       "sync"
       "log"
)

func main() {
       m := sync.Map{}
       
       m.Store("k1", "v1")
       m.Store("k2", "v2")
       
       count := 0
       m.Range(func(key, value interface{}) bool{
              count++
              return true
       })

       log.Println(count)
}

But I didn’t want to range over the items each time I wanted to count them. So I thought of incrementing the counter every time an element was added, and to decrementing it when one was removed. The following is just for practice, I didn’t perform any advanced tests. Continue reading Counting number of items in a concurrent map

Go HTTP server duplicate handler call

It took me a minute or two to figure out why, while doing some experiments, for the following built-in HTTP server use case, the handler was called two times, and so printing the log message two times.

package main

import (
       "log"
       "net/http"
)

func main() {
       http.HandleFunc("/", func(w http.ResponseWriter, r *http.Request) {
              log.Println("call")
       })

       http.ListenAndServe(":8800", nil)
}

The path “/” catches all requests, thus it catches the “/favicon.ico” request sent by the browser. So I’ve just used another path.

Go web app with all assets embedded into the binary

From my first interactions with Go, a very small web app came out. There were several microservices that were exposing data through REST and I needed to quickly browse all of the content in a user friendly way.

I wanted something fast, simple, and easy to integrate new microservices on in the future, as it was the first Go app in the company. I quickly wrote it, deployed it, and the job was done.

To make the deploy process very easy, I wanted a one file application, so I used the go-bindata package to embed the HTML files into the binary, and the CSS and JS files (jQuery and Bootstrap) were served from the official CDNs (now, to show a full example, I’ve embedded all of them).

Take a look at the Micro UI source code.

From PHP to Go

Besides being very powerful, Go is a clean language. It was easy to get started with it, despite it has some obvious major differences if coming from a language like PHP. I knew some of them, cause they’re specific to any compiled language, while in an interpreted one you have to work in order to get their benefits.

I got comfortable with them and even wished PHP had them. I’m not comparing the two languages by considering one to be better or worse, I’m only telling some differences that caught my eye, even if they are normal to be.

It got me happy about writing code in a very different way, putting aside some things that are normal in other contexts and I was used to. Continue reading From PHP to Go

Go concurrency is elegant and simple

These days I wanted to speed up some data retrieval with Go. Its concurrency model is elegant and simple, it has everything you need built-in.

Let’s say there are some articles that need to be fetched from an API. I have the IDSs of all the articles, and I can fetch them one by one. One request can take even a second, so I added a 1 second sleep to simulate this.

type Article struct {
       ID    uint
       Title string
}

func GetArticle(ID uint) Article {
       time.Sleep(time.Second * 1)
       return Article{ID, fmt.Sprintf("Title %d", ID)}
}

The classic way of doing this is making a request for each article, wait for it to finish, store the data.

var articles []Article
var id uint

for id = 1; id <= 10; id++ {
       log.Println(fmt.Sprintf("Fetching article %d...", id))
       article := GetArticle(id)
       articles = append(articles, article)
}

log.Println(articles)

With a 1 second response time it takes 10 seconds. Now imagine 100 articles or more. Continue reading Go concurrency is elegant and simple

Pipelines and workers in Go

I have some lists of users that I get and, for each user, I need to apply some rules (text formatting, max length, and who knows what other business rules can come up in the future), then send it further to another service. If I get the user again, I have to ignore them from the entire process. If one of the rules tells the user is not eligible, I have to stop the entire process, no need to go the next rules.

If you read the previous paragraph again, you can see some if statements that you should avoid from the technical implementation, but of course, not from the business rules:

  • If user was already processed, continue
  • If max length is exceeded, truncate
  • If a rule tells user is not OK, stop

I look at the rules as being some workers in a pipeline. Every worker does its job and sends its work to the next worker. Here’s how I’ve handled this. Continue reading Pipelines and workers in Go